Table 2

Adjusted marginal estimated means of county uninsured rates by Medicaid expansion and Diabetes Belt status, income ≤138% of FPL, 2012 through2016

County typeYearATE* (95% CI)
20122013201420152016
Diabetes Belt
 Nonexpansion39.538.834.730.327.512.3 (10.9–13.7)
 Expansion37.536.421.115.113.3
Non-Belt
 Nonexpansion43.042.136.832.230.24.9 (4.4–5.3)
 Expansion33.632.524.218.115.9
  • Data are percentages.

  • * The average treatment effect (ATE) was computed as the difference in uninsured rates between 2012 and 2016 between Medicaid expansion and nonexpansion states. We estimated ATEs separately for Diabetes Belt and non-Belt counties. The 95% CIs were corrected for multiple comparisons by using Bonferroni methods. P values are not shown because they were <0.001 for all pairwise comparisons between 2012 and subsequent years, and between adjacent years within county type, and for all pairwise comparisons across county types within the same year after Bonferroni correction. The full model from which these marginal means were derived is shown in Supplementary Table 2.